System status

Live checks, measured every 10 seconds from the production host. No estimates, no simulated uptime.

Service commitment

• Availability target: 99.9% monthly for the agent API (measured, not claimed).
• Incident handling: acknowledged within 30 minutes during business hours; post-incident note within 48 hours.
• Fallback: model provider failure switches automatically to the secondary provider (verified 401 → 200).
• Maintenance window: daily 04:00 Asia/Riyadh — quality sampling and fleet checks.

Data source: /status-api/status · refresh on page load.